J Econ Interact Coord (2016) 11:151–169DOI 10.1007/s11403-015-0145-9

REGULAR ARTICLE

A case study for a new metrics for economic complexity:The Netherlands

Andrea Zaccaria · Matthieu Cristelli ·Roland Kupers · Andrea Tacchella ·Luciano Pietronero

Received: 19 February 2014 / Accepted: 5 January 2015 / Published online: 29 January 2015© The Author(s) 2015. This article is published with open access at Springerlink.com

Abstract We present a new approach for the economic analysis of countries, whichwe apply to the case of the Netherlands. Our study is based on a novel way to quan-tify exported products’ complexity and countries’ fitness which has been recentlyintroduced in the literature. Adopting a framework in which products are clustered insectors, we compare the different branches of the export of the Netherlands, takinginto account the time evolution of their volumes, complexities and competitivenessesin the years 1995–2010. The High Tech and Life Sciences sectors share high qualityproducts but low competitiveness; the opposite is true for Horticulture and Energy.Weanalyze in detail the Chemicals sector, finding a declining global complexity whichis mostly driven by a shift towards products of lower quality. A growth forecast isalso provided. In light of our results we suggest a differentiation in policy between thecountry’s self-defined industrial sectors.

Keywords Complexity · Competitiveness · Trade · Industry

JEL Classification C6 · 011 · C23

A. Zaccaria (B) · M. Cristelli · A.Tacchella · L. PietroneroISC-CNR, via dei Taurini, 19, 00185 Rome, Italye-mail: and.zaccaria@gmail.com

R. KupersSmith School of Enterprise and the Environment, University of Oxford, Hayes House,75 George Street, Oxford, UK

A. Tacchella · L. PietroneroDip. di Fisica, “Sapienza” University of Rome, P.le A. Moro 2, 00185 Rome, Italy

123

152 A. Zaccaria et al.

1 Introduction

One of the most important fields of research in Economics concerns the study of eco-nomic growth and the analysis of the causes leading to the observed differences ingrowth rates across countries. In the standard economic literature one usually considersthe so called production functions, which connect the observed or predicted growth tosuitable proxies for human and physical capital (Krugman andWells 2013). However,when the correlation between the growth rate and various state variables is studied,mixed results are found (Barro and Sala-i Martin 2004). For example, the presenceof a connection between growth and investment spendings in Research and Devel-opment has been questioned on an empirical basis by Jones (1995). Hausmann andRodrick have proposed a general equilibrium model in which a central role is playedby the learning process of countries, and in particular by their ability to identify whichtechnologies can be better adapted to their situations (Hausmann and Rodrick 2003).

Recently, an approach based on the Science of Complexity (Anderson 1972) havebeen proposed by Hidalgo and Hausmann (2009). These authors claim that their Eco-nomic Complexity Index outperforms various explanatory variables which have beenproposed in the economic literature, such as the Worldwide Governance Indicatorscalculated by the World Bank, the World Economic Forum’s Global CompetitivenessIndex, and various educational measures (see Hausmann et al. 2011 for a detailedanalysis). Hidalgo and Hausmann’s approach has been criticized and deeply revisitedin Tacchella et al. (2012), Caldarelli et al. (2012), Tacchella et al. (2013), Cristelliet al. (2013). The analyses of the present paper are based on this last approach, whichis founded on a new viewpoint on the relation between the countries and the productsthey export. The standard economic approach (Smith 1776; Ricardo 1817; Romer1990; Grossman and Helpman 1991; Flam and Flanders 1991) to analyze the exportsof a country relies on the assumption that the best strategy is to focus on a few, first-rate products. These products would capture a consistent fraction of the availablemarket, assuring high profits and employment. In order to decide in which productsa country should specialize, Ricardo utilized the concept of comparative advantage,showing that if countries produce and trade only some chosen products the systemwill obtain a total improvement. A contrary view, explained for example in Dosi et al.(1990), suggests instead that countries focus on those products in which they havean absolute advantage, which relies mainly on technology.1 In order to explain theempirical evidence for countries’ diversification one can single out the new trade the-ory (Krugman 1979), in which a production function of the form proposed in Dixit andStiglitz (1977) is used. However, as correctly pointed out by Hidalgo and Hausmann(2010), such an approach does not give, in principle, any clue about which productswill be chosen by countries, and in particular by the less diversified ones. In order tostudy this fundamental problem, both combinatorial (Hidalgo and Hausmann 2010)and network-based approaches has been proposed (Hidalgo et al. 2007; Zaccaria et al.2014).

1 We have checked that the main features of our analysis, which rely on the triangular structure of the exportmatrix, are not much dependent on the choice of the specific measure to assess countries’ export baskets.

123

A new metrics for economic complexity: The Netherlands 153

Empirically, the structure of the country-product matrix is peculiar and suggeststhat the assumption of countries’ specialization is not valid. Let us consider the matrixMcp whose entries take the value 1 if the country c exports the product p and 0 oth-erwise. In practice, we computed the Revealed Comparative Advantage as defined byBalassa (1965),2 and we used a threshold value of 1 to make the matrix binary, asdone in Hidalgo and Hausmann (2009) and Tacchella et al. (2012). Let us arrangerows from the most to the less diversified country and columns from the most tothe less ubiquitous product. In a Ricardian world we would expect to find a blockdiagonal matrix in which each country is specialized in a small set of suitable prod-ucts. On the contrary, as shown in Hidalgo and Hausmann (2009), Tacchella et al.(2012), Caldarelli et al. (2012), Mcp turns to be triangular. In particular, the countrieswhich export rare and complex products are, on a first approximation, the ones thatwe would call developed, at least from the point of view of the usually adopted indi-cators; the important point is that these countries are also the most diversified ones.This observation leads to the introduction of a new metrics for products’ complexityand countries’ fitness (Tacchella et al. 2012), which is particularly suitable to studythe country’s competitiveness with respect to the global market. This methodologywill be explained in a more detailed way in the next section.3 In this paper we usethis metrics to study the characteristics of one country in particular, the Netherlands.We believe that this country, among the many that could have been chosen to illus-trate this methodology, can be an interesting test bed for the following reasons: i) TheNetherlands has a well diversified export basket; as a consequence, a broad range ofproducts can be included in our analysis. ii) This country has articulated an explicitpolicy to build economic advantage based on groupings of products and industries intosectors. This raises the question whether our methodology enables a test of the relativecompetitiveness of such self-defined sectors for industrial policy. We find that it doesso, although to the exclusion of services sectors with are not covered by the methodol-ogy. iii) The Netherlands did not experience political and social turmoil, which may, apriori, undermine an export based analysis. iv) Not least, one of the authors (RK) hasa relevant experience with respect to this specific country, among many others, and inparticular he has been involved in consulting about strategic planning and industrialdevelopment.

The paper is organized as follows. First we introduce the methodology to measureour quantities of interest, namely countries’ fitness and products’ complexity, startingfrom the yearly export flows. Then we describe our data and we illustrate the principleof the products’ division in sectors which we will adopt in the rest of the paper. Atthis point we analyze the exports of the Netherlands using the previously introducedframework, with a particular emphasis on the time evolution of volumes and complex-ities, the export spectroscopy and a detailed analysis of a sample sector, Chemicals.

2 For each country and product, the Revealed Comparative Advantage is the ratio of two ratios: the numer-ator is the weight of the product with respect to that country’s basket; the denominator is the weight of thetotal trade of that product with respect to the total world export. This normalization naturally takes intoaccount the country’s size.3 We point out that this coupling between countries’ and products’ evaluations is a fundamental differencebetween this method and other measures of products’ sophistication, as the one proposed by Lall (2006),which is based on exporters’ income.

123

154 A. Zaccaria et al.

The last section draws some conclusions and illustrates some perspectives for futurework.

2 A new metrics for countries’ fitness and products’ complexity

In this section we illustrate a new approach to measure the fitness of countries and thecomplexity of the exported products. This methodology has been recently introducedby Tacchella et al. (2012). The idea is the following. Each country has some capabil-ities, which represent its social, cultural and technological structure [this concept hasbeen discussed in the economic growth framework in Dosi et al. (1990) and Lall et al.(2006)]. The capabilities permit to produce and export products, so the extension of thedistribution of products and their complexity are linked to each country fitness; in par-ticular, the complexity of a product increases with the capabilities which are needed inorder to produce it, and the fitness is ameasure of the complexity and the diversificationof the exported products.4 In order to make this line of reasoning more quantitative,the starting point is the global structure of the previously defined matrix Mcp. Oncecountries and products are suitably arranged, this matrix is triangular, showing thatdeveloped countries have a diversified export, while poor or less developed countriesexport fewer, lower complexity products. This is clear from Fig. 1, in which we showa spectroscopy5 of the Netherlands and three other countries. To obtain these figureswe sort the products by complexity and we plot their volume. We have grouped theproducts into bins of 5 products each on a complexity basis for clarity purposes; wepoint out that the aim of this representation is only to have a single comprehensive viewof the export quality and diversification, and not to analyze economic sectors, whichis done in the following sections. One can see that the Netherlands cover a wide rangeof products and it is not focused on high nor low complexity products. This is whatwe expect from a high-fitness country; in fact, its position in the fitness ranking is 11.Also Japan has an excellent diversification, with a special focus on high complexityproducts, while Nigeria and Qatar export only products whose complexity is quite low,for example petroleum or simple by-products. It is interesting to notice that Qatar hasone of the highest per capita GDPs: this fact is not immediately translated into a highnumber of capabilities. Diversification may lead to an immediate, zero-order estimatefor countries’ fitness as the number of products they export. A natural further step isto calculate the Fitness of a country as the sum of the complexities of the productsit exports. The evaluation of products’ complexity is more subtle. We start noticingthat the fact that a developed country exports a given product gives little informationon the complexity of that product, because developed countries are diversified and sothey export products with different qualities. On the contrary, if a product is exportedby a country with a low fitness score, we will know that it can not be too complex.As a consequence, one has to assign complexities in such a way to weight more the

4 In the following, we will use the export as a proxy for production. We are aware of the roughness of thisapproximation. However, we do not have access to similar databases on internal markets, and we believethat this partial analysis can contribute to the study of countries’ development and growth in a significantway.5 In Physics the spectroscopy is the quantification of the constituents of a physical or chemical compound.

123

A new metrics for economic complexity: The Netherlands 155

Fig. 1 The countries’ spectroscopy shows the distribution of the exported volumes for different products,ordered according to their complexity. One can notice that Netherlands, along with Japan, has not focusedon a single class of products; instead, both low complexity and high complexity products are present intheir export basket. On the contrary, Nigeria and Qatar, whose economies are based on natural resources,have a poorly diversified export

123

156 A. Zaccaria et al.

low-fitness countries. So in the case of products’ complexity a non linear approachmust be adopted. Having defined the fitnesses in terms of the complexities, andvice versa, an iterative algorithm has to be constructed: one starts from an initialguess and lets the two maps evolve. In practice, in order to calculate the fitness Fc ofthe country c, and the complexity Qp of the product p, the following set of non linearcoupled equations has been proposed (Tacchella et al. 2012):

F̃ (n)c =

∑

p

McpQ(n−1)p (1)

Q̃(n)p = 1

∑c Mcp

1F (n−1)c

(2)

F (n)c = F̃ (n)

c

〈F̃ (n)c 〉c

(3)

Q(n)p = Q̃(n)

p

〈Q̃(n)p 〉p

(4)

where the normalization of the intermediate tilded variables is made as a second stepby dividing each F̃ (n)

c and Q̃(n)p by the respective averages,

〈F̃ (n)c 〉c ≡ 1

C

∑

c

F̃ (n)c 〈Q̃(n)

p 〉c ≡ 1

P

∑

p

Q̃(n)p

where C and P are the total number of countries and products in the database, respec-tively, and n is the iteration index. In order to obtain an extensive metrics, which takesinto account also the amount of the exports, one can replace the matrix Mcp withits weighted counterpart Wcp, which is defined as Wcp = qcp∑

c qcpwhere qcp is the

export flow out the country c concerning the product p. We point out that the case ofa monopoly is already included in the algorithm: in this situation the complexity ofthe product is equal to the exporter’s fitness.

The fixed point of these maps has been studied with extensive numerical simu-lations and it is found to be stable and not depending on the initial conditions. Werefer to Cristelli et al. (2013) for an explanation of the features of this approach andfor a detailed comparison with the Method of Reflections proposed in Hidalgo andHausmann (2009). Instead, for a detailed study of the convergence properties of thisnon linear map, see Pugliese et al. (2014).

3 Description of data and sectors

The analyses conducted in this paper are based on the import-export flows as registeredin the UN COMTRADE database, processed by BACI (Gaulier and Zignago 2010).This database includes the exports of 200 countries and coversmore than 5,000 productcategories for a period ranging from 1995 to 2010. The volumes are expressed in the

123

A new metrics for economic complexity: The Netherlands 157

same currency (namely, US Dollars, year 2000) so we can compare different countriesand different years. There are many possible categorizations for products; we will baseour study on the Harmonized System 2007, 4-digit coding, for a total of 1,131 setsof products. After a data cleaning procedure, whose aim is to remove obvious errorsin the database records, the number of countries fluctuates between 146 and 148 overthe years.

We organize these sets of products taking the sectors which have been recently pro-posed by the Dutch Government6 as a starting point. We are aware that these sectorsare in some ways different from the ones usually considered in the economic litera-ture. Since we focus on the Netherlands, we prefer to adopt a partitioning which, inthe opinion of the Dutch Government, is more suitable in order to assess their policy.As the original aim of this categorization was different than ours, we expect to haveto handle with care the placement of products. The original sectors were 9: Agri-food, Horticulture, High-Tech, Energy, Logistics, Creative Industries, Life Sciences,Chemicals, Water. In the following we will consider only physical products, whichdoes not include services. As a consequence, we will not include Logistics and Waterin our analysis. After a first inspection of the database we decide to consider two moresectors: Metals and Other products. Now we briefly describe the sectors, adding someexamples for clarity.

1. AgriFood: The integrated agri-food supply chain. This includes products relatedto food processing, distribution, and retail.

2. Horticulture: All cultivated flowers, vegetables, fruits, plants. There could be apotential overlap with sector 1 in eatable plants, which however turns out to be asmall sector. We have included these products in sector 1.

3. High Tech: Mainly electronics and robotics, advanced materials, IT, semiconduc-tors, printing machinery, radars. Cars, trucks and planes are also included.

4. Energy: Products which are directly used to produce energy. Examples includepetroleum, natural gas, radioactive elements, wind turbines and solar panels.We donot have electricity among our exports, because only physical products are takeninto account. As a counterexample, high tech components suitable for nuclearreactors are put into sector 3.

5. Creative industries: This sector includesmostly services, for example theTVseriesBig Brother, with big royalty streams. The physical products within this sectorincludes only books, movies and related products. The total exported volume,expressed in US dollars, is quite low. We have however considered it for the sakeof completeness.

6. Life sciences: Every product that is related to medicine. Examples include phar-maceutical goods, instruments and appliances used in medical, surgical, dental,veterinary or orthopedic sciences, plus the x-ray machinery.

7. Chemicals: Chemical elements, polymers, fertilizers, pneumatic tires, plastic.8. Metals: Pure metals, without any chemical or industrial process.9. Other: Everything not included in the previous classification. For example, spores

and paper.

6 http://www.top-sectoren.nl.

123

158 A. Zaccaria et al.

As a first check we compute the total exported volume for each sector and we compareit with the value taken from the respective 2011 policy report for each topsector andthe 2012 CBS report (Centraal Bureau voor de Statistiek 2012). We find consistentlylower volume values, but this is due to the generality of sectors’ definition (in fact,different Dutch institutions come to different results); in particular, the CBS reportstates that only 40% of the export is covered by the top sectors. We point out that thetotal export approximately lines up with the CBS report and that we want to use thetop sectors only as a way to allocate the products in logical buckets.

1995 2000 2005 2010Year

150

200

250

300

350

400

Tota

l Exp

ort (

blns

$)

Fig. 2 The total Dutch export, expressed in billion dollars, for the years 1995–2010. One can notice aperiod of stasis followed by a fast growth from 2003. The evident decline in 2009 is due to the recent crisis

1995 2000 2005 2010Year

0

50

100

150

Expo

rt

OtherAgriFoodHorticoltureHighTechEnergyCreativeLife SciencesChemicalsMetals

Fig. 3 Dutch exports as a function of time, now divided into sectors. Most of the sectors follow a similarpattern, stressing the presence of a correlation among them. The High-Tech sector is the main contributorof Dutch export. In contrast, the Creative sectors seems to be unimportant with respect to the others

123

A new metrics for economic complexity: The Netherlands 159

In Fig. 2 we plot the total exported volume, in billion dollars, as a function oftime. Three regimes are present: during the first one, from 1995 to 2002, the exportis substantially stable. From 2003 to 2008 volumes rapidly increase until the recent2009 crisis, which brought Dutch exports to the level of 2006. However, in 2010 aclear sign of recovery is present. This pattern can be approximately noted also if onedivides the export basket into sectors, as in Fig. 3, where we plot the time evolutionof the volumes for each sector. One can easily see that High Tech represents the mainsector of Netherlands’ export, covering slightly less than one third of the total volume.On the contrary the Creative sector, which is mostly covered by royalties, which arenot included in the products we consider, is practically negligible.

4 The Netherlands: a novel approach

4.1 Top products analysis

In Fig. 4 we show a synthetic view of the top Dutch exported products, ordered byvolume. The columns refer to:

1. a short product description, mainly taken from the Baci website;7

2. the 4-digit HS2007 code;3. the product complexity, as defined in Eq. 3;4. the exported volume, expressed in billion dollars;5. the Dutch market share, along with the shares of the world seven top competitors;6. an illustration of the market. Each competitor is represented by a circle, whose

center is givenby its Fitness, seeEq. 4, and its exportedvolume.Nationswhich lie atthe top right are the best competitors, because they are high fitness countries whichexport a lot. The radius of the circle is givenby the competitiveness,measured as theratio between the fitness of the country and the average fitness of the competitors.

One can notice that China, but also USA and Germany (whose code is DEU) are themost “dangerous” competitors for the export of the Netherlands. In fact, often theycover a big part of the same market wherein the Netherlands place its top products.Moreover, these countries share a high level of fitness, which means that they havemany different capabilities and, as a consequence, we expect their presence in themarket not to fall in the short run.

4.2 The new description at a glance

The introduction of a measure of products’ complexity gives us the possibility torepresent a comprehensive view of the Dutch export in a single graph. In Fig. 5 weplot the exported volumes, expressed in US Dollars, and the average complexity of thesectors for the years 1995 and 2010. The average complexity is calculated weightingeach product with its volume:

7 http://www.cepii.fr/anglaisgraph/bdd/baci.htm.

123

160 A. Zaccaria et al.

Fig. 4 A synthesis of the main characteristics of the 6 top products of Dutch export. See text for details

〈Q〉S = 1

Vtot

∑

p∈SVpQp (5)

where Vtot = ∑p∈S Vp. With the symbolic expression

∑p∈S we mean that the sum

is performed only on those products which belong on one of the sectors previously

123

A new metrics for economic complexity: The Netherlands 161

0 0.5 1 1.5 2Complexity

0

10

20

30

40

50Ex

porte

d V

olum

es (b

lns $

) OtherAgrifoodHorticultureHigh TechEnergyCreativeLife SciencesChemicalsMetal

1995

0 0.5 1 1.5 2Complexity

0

50

100

150

Expo

rted

Vol

umes

(bln

s $) Other

AgrifoodHorticultureHigh TechEnergyCreativeLife SciencesChemicalsMetals

2010

Fig. 5 Dutch exports for the years 1995 and 2010. For each sector the weighted average complexity andthe exported volume is plotted

0 1 2 3 4Complexity

0

5

10

15

20

Expo

rted

Vol

umes

(bln

s $) Other

AgrifoodHorticultureHigh TechEnergyCreativeLife SciencesChemicalsMetals

Fig. 6 Volume distributions on the complexity axis for the different sectors for the year 2010

defined; as a consequence, each sector S will be characterized by its weighted averagecomplexity 〈Q〉S .

One can notice that those sectors which heavily rely on raw materials, as Energy,Horticulture and Metals, share a low complexity, while those sectors which are basedon a high level of technology and advanced skills and infrastructures, such as HighTech, Life Sciences and Chemicals, have a high average complexity. A quick compar-ison between the two years shows that the exports grow for all sectors (note that the yaxis has been rescaled). On the contrary, the complexities do not share the same trend:some sectors remain approximatively constant (for example, Energy and High Tech)while others show a remarkable decline, for example Chemicals. We will discuss thisaspect in detail in the next Section.

In order to visualize the sector compositions we show in Fig. 6 how the productsare distributed along the complexity axis. In other words, we break the columns of

123

162 A. Zaccaria et al.

the previous figures in the respective components, preserving a certain aggregation toimprove the readability of the chart. Some sectors, like Energy, are confined to a smallinterval around the average,while others, likeChemicals, cover a broad rangeof values.As we will see in the following Section, this fact leads to important consequences forthe time evolution of the sectors’ complexity.

4.3 Time evolution

In this paragraph we study if and how the features of the Netherlands’ export changedwith time. We start from the simplest analysis, the spectroscopy. In Fig. 7 we showour results for three different years: from top to bottom, 1995, 2002 and 2009. Itis evident that the Netherlands’ main characteristic, the good diversification of theexport, remains unchanged. This is confirmed by a further analysis of the percentiles,which we do not show for reasons of space.

In Fig. 8 we plot the temporal evolution of the sectors’ complexity, as defined inEq. 5. On average, these values are declining. Broadly speaking, this is due to theentrance of low fitness competitors in the market, as expected by the mathematicalexpression of products’ complexity. We note the presence of different patterns in time.For some sectors, for example Agrifood and Chemicals, this trend is constant. Theother sectors do not lose complexity in such a consistent way. Given our definitionof sectors’ complexity as a volume weighted average, we can ask ourselves if thisbehavior is due to the time evolution of the single products’ complexity or to weightsrearrangements inside a given sector. In order to answer to questions like these, oneusually studies how the average 〈Qq1V q2〉 varies with q1 and q2. Clearly, the exponentsregulate the relative weight of the two variables. In particular, for q1 = q2 = 1 werecover our definition of weighted average complexity, while for q1 = 0 and q2 = 1we have, up to normalization constants, the same behavior of Fig. 3. In this case onlythe volumes count, and we see that in the first half of the considered time interval theexport is constant and then it increases. On the contrary, if we consider the case inwhich q1 = 1 and q2 = 0 we find the average complexity in which each exportedproduct has the same weight. Taking into account the Chemicals sector, we show inFig. 9 that this quantity is declining in amuch less pronounced way. This canmean thateither the Netherlands are focused on products which have lost complexity throughthe years, or that this country has changed its basket from high complexity to lowcomplexity products. In any case, this trending is worrying because it underlines adeterioration of exports’ quality, in contrast with the behavior of volumes, which areincreasing.8

In order to confirm this hypothesis quantitatively,we studied the correlation betweenvolumes and complexities inside the Chemicals sector. The Pearson coefficient for1995 is equal to 0.026 and it grows to 0.126 in 1998, a weak but positive correlation(that is, good products are exported more). Then it starts declining: for the year 2002we find 0.03 and for 2007 −0.048. For the last year we studied, 2010, the correlation

8 In a similar way we have checked the behavior of the Agrifood sector, finding analogous results.

123

A new metrics for economic complexity: The Netherlands 163

Fig. 7 The spectroscopy of the Netherlands for three different years: from top to bottom, 1995, 2002 and2009. The country shows an excellent diversification of its products, with no substantial modification overthe years

between volumes and complexities decreases to −0.108.9 This behavior evidencesthat in the last years the good products have lost weight in the Chemicals sectorbecause, on the contrary, a situation in which all the products had lost complexity

9 Being the sample size very high (about 200) the correlations are statistically significant for the extremevalues. For example, for the 1998 we have a p value of 0.078 for a two tailed test and 0.039 for a one tailedtest. Moreover, we observe a declining trend during the years.

123

164 A. Zaccaria et al.

1995 2000 2005 2010Year

0

0.5

1

1.5

2C

ompl

exity

OtherAgrifoodHorticultureHigh TechEnergyCreativeLife SciencesChemicalsMetals

Fig. 8 Sectors’ complexity as functions of time. Most of them remain roughly constant, while others, likeChemicals, show a coherent decline over the years

1995 2000 2005 2010Year

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

Com

plex

ity

Standard AverageWeighted average

Chemicals

Fig. 9 The time evolution of the simple and weighted average complexity for the Chemicals sector. Thelatter shows a more remarkable decline

keeping their volumes constant would have led to a constant correlation coefficient.As a consequence, we can state that the complexity loss of this sector has been causedby a weight rearrangement towards lower complexity products over the years.

In order to estimate the growth of the Netherlands we have studied how the othercountries evolve, as trajectories in the Fitness-GDP per capita plane, using a recentlyintroduced methodology, which is called Selective Predictability Scheme (SPS) (M.

123

A new metrics for economic complexity: The Netherlands 165

Fig. 10 The results of the SPS analysis predict a substantial stasis of both fitness and GDP per capita inthe next 10years

Cristelli, A. Tacchella and L. Pietronero, HeterogeneousDynamics of Economic Com-plexity, Submitted). The basic assumption is that similar initial conditions lead tosimilar dynamics, resembling the method of analogues, well known in the physics ofdynamical systems. The results of our analysis are depicted in Fig. 10. First of all,we have a good predictability. In order to quantify the goodness of our forecasts weuse a measure of the future concentration of those points which now are close to theNetherlands. These points, in the past, moved from the area that the Netherlands iscrossing to the other areas with a frequency given by the colors. Having the darkerareas the same fitness and GDP per capita of 2012, we predict that Netherlands willnot considerably grow in the next ten years.

5 Comparison with other countries and competitiveness

In Fig. 11 we compare the fitness rankings of various countries with the one of theNetherlands. To compare countries we use the rankings instead of the numerical valuesbecause of the algorithm normalization. Indeed, all Western countries are sharinga declining trend of the fitness. This is due to the normalization procedure of theiterations, in which the average fitness is kept constant: as a consequence, if the fitnessof the emerging countries increases, the one of the developed countries must decrease.The fitness of theNetherlands is almost constant in the years 1995–2010, revealing thatthis country not only has preserved the diversification of its basket of export productsbut also saved it from the emergence of new competitors. Figure 11 gives a rankings-based comparison which gives, in addition, some insights about other countries.While

123

166 A. Zaccaria et al.

Fig. 11 Time evolution of thefitness for the Netherlands(magenta). Also some Westerncompetitors and the BRICcountries (green) are highlighted(color figure online)

Year1996 1998 2000 2002 2004 2006 2008 2010

Ran

king

(Fitn

ess)

50

40

30

20

10

1

BRA

RUS

CHI

IND

DEU

USA

BELNLD

Germany, the Netherlands, and Belgium show a constant behavior, USA experienceda declining trend in the last 5 years. If one considers the patterns of the so calledBRIC countries (green lines), it is evident that our methodology splits them in aclear way. Already in 1995 these countries had very different fitness rankings. Afteran indefinite 5 years phase, in which only India appears to grow, these countriesdiverge. In particular, while China rises in an impressive way, India remains constantin ranking.On the contrary, Brazil andRussia show a remarkable decline, probably dueto an excessive dependence of the export on the primary sector, which has underminedthe possible development of other, more complex sectors. Nowadays the presence ofdifferent behaviors among the BRICs is a well known fact among the experts; using

123

A new metrics for economic complexity: The Netherlands 167

Fig. 12 The time evolution of the sectors’ competitiveness. See the text for details

our methodology one can see that this is a result of a trend which started more than adecade ago.

Now we address the issue of quantifying the competitiveness of the Netherlandswith respect to other countries, across the different sectors.

Even if, as we have seen in the previous sections, the process bywhich the complex-ities are calculated already takes into account the export flows of the other countries,we would like to have a quantification of the importance of the direct competitors ofthe Netherlands which uses the final results of the iteration procedure. A first proposalis given by the following simple formula. Given the product p exported by the countryc, the competitiveness is:

Fc〈Fc〉competi tors

(6)

where the denominator is an average over the competitors, that is, those countrieswhose export of the product p is above a given threshold. We are aware of the rough-ness of this approximation: for example, low fitness countries can be very competitivethanks to their low labor cost. However, we believe that a sector without high fit-ness competitors should be, in principle, richer in opportunity than an full one (Kimand Mauborgne 2005). In Fig. 12 we show how the competitiveness of the sectorsevolves with time. The top Dutch sectors from a competitiveness point of view areEnergy and Horticulture: the fitness of the competitors is, on average, 4 o 10 timeslower than the one of the Netherlands. Despite the low complexity score of thesetwo sectors (see Fig. 8), they have less competitors in the market and so we canstate that the Netherlands could focus on these two sector with confidence. At theopposite, the Chemicals sector, which has a high but quickly declining complex-ity, has a very low competitiveness; we can not recommend to invest only on thissector.

123

168 A. Zaccaria et al.

6 Conclusion and perspectives

The aim of this paper is to illustrate the usefulness of some recently introduced instru-ments to describe and quantify the competitiveness of countries’ economy. In particu-lar, the new framework is able to quantify the fitness of a country, that is, its economic,cultural and technological state and the complexity of its exported products, defined interms of the capabilities needed to prepare or fabricate them. These two quantities arelinked by a set of equations which permit a novel measure of these features and, as aconsequence, a new point of view fromwhich countries and products can be profitablystudied. We applied this methodology to a specific case: the Netherlands. As a firstresult, from the spectroscopy of the country we can say that the Netherlands show agood diversification, which remains stable through the years. In order to give moreprecise insights we clustered the products into nine sectors, which we have studiedseparately. Considering the complexity of the products which make up the sectors, onefinds a heterogeneous situation: for example, the Energy sector is primarily made upby low complexity products; on the contrary, theHigh Tech sector’smain contributionscome from high complexity exports. Other sectors, like Chemicals, are not focusedon a single class of products. From a time evolution perspective we have found thatwhile the growth of the exports can be considered as a common feature of all sectors,for some sectors complexity is declining. We analyzed in detail the case of Chemicals,whose decline has been due to a change of focus towards low complexity products. Afirst study of sectors’ competitiveness suggests that most advantage lies in the Energyand Horticulture sectors, but taking into account that their success is rooted in thehighly diverse economic system that supports them. We suggest that our approach,being based on a novel methodology, is a useful complement to the standard approachbased on relative export volume ranking.

Being aware of the weight of services in the exports of the Netherlands, we planto include them a future analysis. Moreover, also imports have been excluded. Giventhe massive trading position of the Netherlands, we plan to include a discussion ofthe flow-through volumes and of their added values. For the moment, we believe thatour data, which contains the export volumes of all the physical products, can give asufficiently good view of the global situation of this country. We expect, for example,to find an high degree of diversification also for services.

Acknowledgments We thank Riccardo Di Clemente for providing the artwork for Fig. 11. This workwas supported by the European project FET-Open GROWTHCOM (Grant Number 611272) and the ItalianPNR project CRISIS-Lab.

Open Access This article is distributed under the terms of the Creative Commons Attribution Licensewhich permits any use, distribution, and reproduction in any medium, provided the original author(s) andthe source are credited.

References

Anderson PW (1972) More is different. Science 177(4047):393–396Balassa B (1965) Trade liberalisation and “revealed” comparative advantage. Manch Sch 33(2):99–123.

Wiley Online Library

123

A new metrics for economic complexity: The Netherlands 169

Barro RJ, Sala-i Martin X (2004) Economic growth. MIT Press, CambridgeCaldarelli G, Cristelli M, Gabrielli A, Pietronero L, Scala A, Tacchella A (2012) A network analysis of

countries’ export flows: firm grounds for the building blocks of the economy. PloS One 7(10):e47278Centraal Bureau voor de Statistiek (2012) Monitor Topsectoren - Uitkomsten eerste meting, http://tinyurl.

com/lhwuzjn (in Dutch). Accessed 7 Feb 2014. Technical reportCristelli M, Gabrielli A, Tacchella A, Caldarelli G, Pietronero L (2013)Measuring the intangibles: a metrics

for the economic complexity of countries and products. PLoS One 8:e70726Dixit AK, Stiglitz JE (1977) Monopolistic competition and optimum product diversity. Am Econ Rev

67:297–308Dosi G, Pavitt K, Soete LL (1990) The economics of technical change and international trade. LEM Book

SeriesFlam H, Flanders M (1991) Heckscher–Ohlin trade theory. MIT Press, CambridgeGaulier G, Zignago S (2010) Baci: international trade database at the product-level. Centre d’Etudes

Prospectives et d’Informations Internationales http://www.cepii.fr/anglaisgraph/workpap/pdf/2010/wp2010-23

Grossman G, Helpman E (1991) Quality ladders in the theory of growth. Rev Econ Stud 58:43–61Hausmann R, Rodrick D (2003) Economic development as self-discovery. J Dev Econ 72:603–633Hausmann R, Hidalgo CA, Bustos A, Coscia M, Chung S, Jimenez J, Simoes A, Yildirim M (2011) The

atlas of economic complexity. Puritan Press, CambridgeHidalgo CA, Klinger B, Barabási AL, Hausmann R (2007) The product space conditions the development

of nations. Science 317(5837):482–487Hidalgo C, Hausmann R (2009) The building blocks of economic complexity. Proc Natl Acad Sci USA

106:10570–10575Hidalgo C, Hausmann R (2010) Country diversification, product ubiquity, and economic divergence. John

F. Kennedy School of Government, Harvard UniversityJones CI (1995) R&D-based models of economic growth. J Polit Econ 103(4):759Kim WC, Mauborgne R (2005) Blue ocean strategy: how to create uncontested market space and make

competition irrelevant. Harvard Business Press, CambridgeKrugman P (1979) Increasing returns, monopolistic competition, and international trade. J Int Econ

9(4):469–479Krugman P, Wells R (2013) Macroeconomics. Worth Publishers, New YorkLall S, Weiss J, Zhang J (2006) The sophistication of exports: a new trade measure. World Dev 34(2):222–

237Lall S (2006) The technological structure and performance of developing country manufactured exports,

1985–1998. Oxf Dev Stud 28:337–369Pugliese E, Zaccaria A, Pietronero L (2014) On the convergence of the fitness-complexity algorithm

arXiv:1410.0249Ricardo D (1817) On the principles of political economy and taxation. John MurrayRomer P (1990) Endogenous technological change. J Polit Econ 98:71–102Smith A (1776) The wealth of nations. Strahan and CadellTacchella A, Cristelli M, Caldarelli G, Gabrielli A, Pietronero L (2012) A newmetrics for countries’ fitness

and products’ complexity. Sci Rep 2:723Tacchella A, Cristelli M, Caldarelli G, Gabrielli A, Pietronero L (2013) Economic complexity: conceptual

grounding of a new metrics for global competitiveness. J Econ Dyn Control 37(8):1683–1691ZaccariaA,CristelliM,TacchellaA, PietroneroL (2014)How the taxonomyof products drives the economic

development of countries. PLoS One 9(12):e113770

123